Biology subjects

Rosen, G.

Publications and source records attributed to Rosen, G..

12 recordsLinked to original sources

bioRxiv2026

Nevermore: Target-Conditioned Protein-Ligand Representation Learning for Multi-Objective Lead Optimization with Database-Grounded Retrieval

Target-conditioned molecular design requires optimizing binding affinity to a proposed therapeutic protein target while balancing competing developability constraints (e.g., absorption, distribution, metabolism, excretion, and toxicity; ADMET). Yet many computational pipelines either optimize a single objective or rely on fully de novo generation that can be difficult to control and interpret. We present Nevermore, a target-conditioned, database-grounded framework that combines a geometry-aware protein-ligand affinity oracle with Pareto-aware multi-objective search over an explicit molecular feature space. A central design choice is to optimize in count-based Morgan fingerprint space, where each feature corresponds to a chemically meaningful substructure count, enabling discrete, interpretable "bucket-level" edits. Nevermore learns target-conditioned scores by aligning protein and ligand representations under contrastive objectives and using a similarity-based prediction head; the resulting affinity oracle improves over previously reported benchmark baselines, providing a stronger scoring signal for downstream optimization. Nevermore then steers candidate selection by proposing sparse fingerprint edits, re-ranking candidates under multiple objectives, and projecting edited fingerprints back to valid molecules via nearest-neighbor retrieval from a large compound library. This yields efficient screening without exhaustive enumeration and provides transparent attributions that connect optimization steps to concrete chemical motifs. We evaluate Nevermore on two target case studies (Menin and SARS-CoV-2 Mpro). Across targets, the closed-loop search consistently retrieves candidate sets with improved affinity-property trade-offs compared with random sampling and similarity-only retrieval baselines, while maintaining explicit control and interpretability through discrete feature-space edits. These results support database-grounded, feature-space steering as a practical route to target-conditioned multi-objective lead refinement without relying on fully de novo generation.

bioinformatics↗

bioRxiv2025

Transcriptional Profiling of Commonly Used Liver Cancer Cell Lines Reveals Disease-Specific Modeling Potential and Authentication Concerns

Cell lines are essential tools for liver cancer research, yet their molecular fidelity to primary tumors remains incompletely characterized. Here we comprehensively evaluated transcriptomic similarities between commonly used liver cancer cell lines and primary tumor subtypes to guide optimal model selection. We analyzed RNA sequencing data from 541 samples spanning primary HCC, HPBL, CHOL, and FLC tumors, alongside 21 liver cancer cell lines and primary human hepatocytes. Through systematic variance analysis, we identified 2,523 highly variable genes distinguishing cancer subtypes and cell lines, then performed correlation analyses, unsupervised clustering, and pathway enrichment to assess molecular similarities. Molecular subtypes within each cancer type were identified through hierarchical clustering and characterized using pathway analysis. HepG2 cells showed strongest correlation with HPBL (r=0.62), confirming their hepatoblastoma origin despite frequent HCC misclassification. This correlation was driven by shared Wnt pathway dysregulation signatures. Huh7 cells best represented HCC, particularly the immune-modulatory, MYC-activated subtype with the highest median correlation. RBE cells optimally modeled CHOL, specifically the dedifferentiated, immune-evasive subtype. Several commonly used cell lines (LO2, SMMC-7721, MHCC97L) and specific publicly available samples demonstrated likely HeLa contamination. Primary human hepatocytes cultured under physioxic conditions better preserved liver-specific transcriptional programs compared to standard culture. No established cell line analyzed represented FLC strongly, identifying the need for a standard, available model. This transcriptomic framework provides evidence-based guidance for selecting appropriate liver cancer cell line models and highlights the critical need for rigorous cell line validation to improve experimental design and translational relevance of liver cancer research.

cancer biology↗

bioRxiv2025

The Naïve Bayes Classifier as an Out-of-Distribution Detector of Novel Taxa

Detecting sequences from novel taxa remains a key challenge in metagenomic classification, as reference databases rarely capture the full extent of microbial diversity. We investigate the Naive Bayes Classifier++ (NBC++) as an out-of-distribution (OOD) detector by analyzing its log-likelihood scores across simulated and real metagenomic datasets. By partitioning reference databases and introducing taxonomic novelty, we derive thresholds that distinguish known from unknown reads at multiple taxonomic levels. These thresholds remain consistent across database sizes, indicating that once a lineage is represented, novelty detection performance stabilizes. Applied to a human gut metagenome, the thresholds reflect differences in database density and classification confidence. This work characterizes how NBC++ responds to novelty and illustrates its use in evaluating unclassified metagenomic reads.

bioinformatics↗

bioRxiv2024

Scorpio : Enhancing Embeddings to Improve Downstream Analysis of DNA sequences

Analysis of genomic and metagenomic sequences is inherently more challenging than that of amino acid sequences due to the higher divergence among evolutionarily related nucleotide sequences, variable k-mer and codon usage within and among genomes of diverse species, and poorly understood selective constraints. We introduce Scorpio, a versatile framework designed for nucleotide sequences that employs contrastive learning to improve embeddings. By leveraging pre-trained genomic language models and k-mer frequency embeddings, Scorpio demonstrates competitive performance in diverse applications, including taxonomic and gene classification, antimicrobial resistance (AMR) gene identification, and promoter detection. A key strength of Scorpio is its ability to generalize to novel DNA sequences and taxa, addressing a significant limitation of alignment-based methods. Scorpio has been tested on multiple datasets with DNA sequences of varying lengths (long and short) and shows robust inference capabilities. Additionally, we provide an analysis of the biological information underlying this representation, including correlations between codon adaptation index as a gene expression factor, sequence similarity, and taxonomy, as well as the functional and structural information of genes.

genomics↗

bioRxiv2023

Streamlining Computational Fragment-Based Drug Discovery through Evolutionary Optimization Informed by Ligand-Based Virtual Prescreening

Recent advances in computational methods provide the promise of dramatically accelerating drug discovery. While math-ematical modeling and machine learning have become vital in predicting drug-target interactions and properties, there is untapped potential in computational drug discovery due to the vast and complex chemical space. This paper advances a novel computational fragment-based drug discovery (FBDD) method called Fragment Databases from Screened Ligands Drug Discovery (FDSL-DD), which aims to streamline drug design by applying a two-stage optimization process. In this ap-proach, in silico screening identifies ligands from a vast library, which are then fragmentized while attaching specific at-tributes based on predicted binding affinity and interaction with the target sub-domain. This process both shrinks the search space and focuses on promising regions within it. The first optimization stage assembles these fragments into larger com-pounds using evolutionary strategies, and the second stage iteratively refines resulting compounds for enhanced bioac-tivity. The methodology is validated across three diverse protein targets involved in human solid cancers, bacterial antimi-crobial resistance, and SARS-CoV-2 viral entry, demonstrating the approachs broad applicability. Using the proposed FDSL-DD and two-stage optimization approach yields high-affinity ligand candidates more efficiently than other state-of-the-art computational methods. Furthermore, a multiobjective optimization method is presented that accounts for druglikeness while still producing potential candidate ligands with high binding affinity. Overall, the results demonstrate that integrat-ing detailed chemical information with a constrained search framework can markedly optimize the initial drug discovery process, offering a more precise and efficient route to developing new therapeutics.

bioinformatics↗

bioRxiv2021

Profiling chromatin accessibility in formalin-fixed paraffin-embedded samples

Archived formalin-fixed paraffin-embedded (FFPE) samples are the global standard format for preservation of the majority of biopsies in both basic research and translational cancer studies, and profiling chromatin accessibility in the archived FFPE tissues is fundamental to understanding gene regulation. Accurate mapping of chromatin accessibility from FFPE specimens is challenging because of the high degree of DNA damage. Here, we first showed that standard ATAC-seq can be applied to purified FFPE nuclei but yields lower library complexity and a smaller proportion of long DNA fragments. We then present FFPE-ATAC, the first highly sensitive method for decoding chromatin accessibility in FFPE tissues that combines Tn5-mediated transposition and T7 in vitro transcription. The FFPE-ATAC generates high-quality chromatin accessibility profiles with 500 nuclei from a single FFPE tissue section, enables the dissection of chromatin profiles from the regions of interest with the aid of hematoxylin and eosin (H&E) staining, and reveals disease-associated chromatin regulation from the human colorectal cancer FFPE tissue archived for more than 10 years. In summary, the approach allows decoding of the chromatin states that regulate gene expression in archival FFPE tissues, thereby permitting investigators, to better understand epigenetic regulation in cancer and precision medicine.

genomics↗

bioRxiv2021

MetaMutationalSigs: Comparison of mutational signature refitting results made easy

SummaryThe analysis of mutational signatures is becoming increasingly common in cancer genetics, with emerging implications in cancer evolution, classification, treatment decision and prognosis. Recently, several packages have been developed for mutational signature analysis, with each using different methodology and yielding significantly different results. Because of the nontrivial differences in tools refitting results, researchers may desire to survey and compare the available tools, in order to objectively evaluate the results for their specific research question, such as which mutational signatures are prevalent in different cancer types. There is a need for a software that can aggregate results from different refitting packages and present them in a user-friendly way to facilitate effective comparison of mutational signatures. Availability and implementationMetaMutationalSigs is implemented using R and python and is available for installation using Docker and available at: https://github.com/EESI/MetaMutationalSigs ContactGail Rosen (glr26@drexel.edu). Supplementary informationMore information about the package including test data and results are available at https://github.com/EESI/MetaMutationalSigs

bioinformatics↗

bioRxiv2020

Uncovering the Structure and Function of Microbial Communities Formed During Periodic Tilling of TNT and DNT Co-Contaminated Soils

1.Environmental contamination by 2,4,6-trinitrotoluene (TNT), historically the most widely used secondary explosive, is a long-standing problem in former military conflict areas and at manufacturing and decommissioning plants. In field test plots at a former explosives manufacturing site, removal of TNT and dinitrotoluenes (DNTs) was observed following periods of tillage. Since tilling of soils has previously been shown to alter the microbial community, this study was aimed at understanding how the microbial community is altered in soils with historical contamination of nitro explosives from the former Barksdale TNT plant. Samples of untilled pristine soils, untilled TNT-contaminated soils and tilled TNT-contaminated soils were subjected to targeted amplicon sequencing of 16S ribosomal RNA genes in order to compare the structure of their bacterial communities. In addition, metagenomic data generated from the TNT tilled soil was used to understand the potential functions of the bacterial community relevant to nitroaromatic degradation. While the biodiversity dropped and the Burkholderiales order became dominant in both tilled and untilled soil regardless of tillage, the bacterial community composition at finer taxonomic levels revealed a greater difference between the two treatments. Functional analysis of metagenome assembled genome (MAG) bins through systematic review of commonly proposed DNT and TNT biotransformation pathways suggested that both aerobic and anaerobic degradation pathways were present. A proposed pathway that considers both aerobic and anaerobic steps in the degradation of TNT in the scenario of the tilled contaminated soils is presented. 2. ImportanceIn this study, TNT and DNT removal has been observed in field-scale experiments following periodic tilling of historically contaminated soils. The microbial community structures of uncontaminated pristine soils, untilled contaminated soils, and tilled contaminated soils were investigated using high-throughput sequencing platforms. In addition, shotgun metagenome libraries of samples from tilled contaminated soils were generated. The results indicated that a significant shift of the bacterial community at the family level between tilled and untilled contaminated soils, with tilled soils being dominated by Alcaligenaceae and untilled soils by Burkholderiacea. In-depth metagenomic analysis of samples from tilled contaminated soils, indicate the presence of genes that encode for enzymes that potentially could lead to mineralization of TNT and DNT under mixed aerobic and anaerobic periods.

genomics↗

bioRxiv2020

Learning, Visualizing and Exploring 16S rRNA Structure Using an Attention-based Deep Neural Network

Recurrent neural networks (RNNs) with memory (e.g. LSTMs) and attention mechanisms are widely used in natural language processing because they can capture short and long term sequential information for diverse tasks. We propose an integrated deep learning model for microbial DNA sequence data, which exploits convolutional networks, recurrent neural networks, and attention mechanisms to perform sample-associated attribute prediction--phenotype prediction--and extract interesting features, such as informative taxa and predictive k-mer context. In this paper, we develop this novel deep learning approach and evaluate its application to amplicon sequences. We focus on typically short DNA reads of 16s ribosomal RNA (rRNA) marker genes, which identify the heterogeneity of a microbial community sample. Our deep learning approach enables sample-level attribute and taxonomic prediction, with the aim of aiding biological research and supporting medical diagnosis. We demonstrate that our implementation of a novel attention-based deep network architecture, Read2Pheno, achieves read-level phenotypic prediction and, in turn, that aggregating read-level information can robustly predict microbial community properties, host phenotype, and taxonomic classification, with performance comparable to conventional approaches. Most importantly, as a further result of the training process, the network architecture will encode sequences (reads) into dense, meaningful representations: learned embedded vectors output on the intermediate layer of the network model, which can provide biological insight when visualized. Finally, we demonstrate that a model with an attention layer can automatically identify informative regions in sequences/reads which are particularly informative for classification tasks. An implementation of the attention-based deep learning network is available at https://github.com/EESI/sequence_attention.

bioinformatics↗

bioRxiv2020

RRM2B is frequently amplified across multiple tumor types: non-oncogenic addiction and therapeutic opportunities

RRM2B plays a crucial role in DNA replication, repair and oxidative stress. While germline RRM2B mutations have been implicated in mitochondrial disorders, its relevance to cancer has not been established. Here, using TCGA data, we investigated RRM2B alterations in cancer. We found that RRM2B is highly amplified in multiple tumor types, particularly in MYC-amplified tumors, and is associated with increased RRM2B mRNA expression. We also observed that the chromosomal region 8q22.3-8q24, is amplified in multiple tumors, and includes RRM2B, MYC along with several other cancer-associated genes. An analysis of genes within this 8q-amplicon showed that cases that have both RRM2B-amplified along with MYC have a distinct pattern of amplification compared to unaltered cases or cases that have amplifications in RRM2B or MYC only. These other 8q-proteins were shown to interact functionally within the RRM2B network of DNA repair, hypoxia and apoptosis regulating proteins. Notably, RRM2B-amplified tumors are characterized by mutation signatures of defective DNA repair and oxidative stress, and in some cancers also associated with poor clinical outcome. These findings suggest that some cancers may require RRM2B for cellular survival, providing novel therapeutic opportunities in these cancers.

genomics↗

bioRxiv2020

Characterizing geographical and temporal dynamics of novel coronavirus SARS-CoV-2 using informative subtype markers

We propose an efficient framework for genetic subtyping of SARS-CoV-2, the novel coronavirus that causes the COVID-19 pandemic. Efficient viral subtyping enables visualization and modeling of the geographic distribution and temporal dynamics of disease spread. Subtyping thereby advances the development of effective containment strategies and, potentially, therapeutic and vaccine strategies. However, identifying viral subtypes in real-time is challenging: SARS-CoV-2 is a novel virus, and the pandemic is rapidly expanding. Viral subtypes may be difficult to detect due to rapid evolution; founder effects are more significant than selection pressure; and the clustering threshold for subtyping is not standardized. We propose to identify mutational signatures of available SARS-CoV-2 sequences using a population-based approach: an entropy measure followed by frequency analysis. These signatures, Informative Subtype Markers (ISMs), define a compact set of nucleotide sites that characterize the most variable (and thus most informative) positions in the viral genomes sequenced from different individuals. Through ISM compression, we find that certain distant nucleotide variants covary, including non-coding and ORF1ab sites covarying with the D614G spike protein mutation which has become increasingly prevalent as the pandemic has spread. ISMs are also useful for downstream analyses, such as spatiotemporal visualization of viral dynamics. By analyzing sequence data available in the GISAID database, we validate the utility of ISM-based subtyping by comparing spatiotemporal analyses using ISMs to epidemiological studies of viral transmission in Asia, Europe, and the United States. In addition, we show the relationship of ISMs to phylogenetic reconstructions of SARS-CoV-2 evolution, and therefore, ISMs can play an important complementary role to phylogenetic tree-based analysis, such as is done in the Nextstrain [1] project. The developed pipeline dynamically generates ISMs for newly added SARS-CoV-2 sequences and updates the visualization of pandemic spatiotemporal dynamics, and is available on Github at https://github.com/EESI/ISM and via an interactive website at https://covid19-ism.coe.drexel.edu/. Author SummaryThe novel coronavirus responsible for COVID-19, SARS-CoV-2, expanded to reportedly 8.7 million confirmed cases worldwide by June 21, 2020. The global SARS-CoV-2 pandemic highlights the importance of tracking viral transmission dynamics in real-time. Through June 2020, researchers have obtained genetic sequences of SARS-CoV-2 from over 47,000 samples from infected individuals worldwide. Since the virus readily mutates, each sequence of an infected individual contains useful information linked to the individuals exposure location and sample date. But, there are over 30,000 bases in the full SARS-CoV-2 genome--so tracking genetic variants on a whole-sequence basis becomes unwieldy. We describe a method to instead efficiently identify and label genetic variants, or "subtypes" of SARS-CoV-2. Applying this method results in a compact, 11 base-long compressed label, called an Informative Subtype Marker or "ISM". We define viral subtypes for each ISM, and show how regional distribution of subtypes track the progress of the pandemic. Major findings include (1) covarying nucleotides with the spike protein which has spread rapidly and (2) tracking emergence of a local subtype across the United States connected to Asia and distinct from the outbreak in New York, which is found to be connected to Europe.

genomics↗

bioRxiv2017

Comprehensive Benchmarking and Ensemble Approaches for Metagenomic Classifiers

BackgroundOne of the main challenges in metagenomics is the identification of microorganisms in clinical and environmental samples. While an extensive and heterogeneous set of computational tools is available to classify microorganisms using whole genome shotgun sequencing data, comprehensive comparisons of these methods are limited. In this study, we use the largest (n=35) to date set of laboratory-generated and simulated controls across 846 species to evaluate the performance of eleven metagenomics classifiers. We also assess the effects of filtering and combining tools to reduce the number of false positives.\n\nResultsTools were characterized on the basis of their ability to (1) identify taxa at the genus, species, and strain levels, (2) quantify relative abundance measures of taxa, and (3) classify individual reads to the species level. Strikingly, the number of species identified by the eleven tools can differ by over three orders of magnitude on the same datasets. However, various strategies can ameliorate taxonomic misclassification, including abundance filtering, ensemble approaches, and tool intersection. Indeed, leveraging tools with different heuristics is beneficial for improved precision. Nevertheless, these strategies were often insufficient to completely eliminate false positives from environmental samples, which are especially important where they concern medically relevant species and where customized tools may be required.\n\nConclusionsThe results of this study provide positive controls, titrated standards, and a guide for selecting tools for metagenomic analyses by comparing ranges of precision and recall. We show that proper experimental design and analysis parameters, including depth of sequencing, choice of classifier or classifiers, database size, and filtering, can reduce false positives, provide greater resolution of species in complex metagenomic samples, and improve the interpretation of results.

genomics↗